Comparing How Commercial Different Search Categories are for Search Advertising
Search engines collect a lot of information from their users, much of which they don’t share with us. Some of it would be pretty interesting to see.
A Yahoo patent application describes some information that many who advertise on search engines would probably enjoy seeing very much, something that Yahoo refers to as a Search Category Commercialization Index (SCCI).
A quick aside, one of the things that I like about looking at patent filings is that they sometimes provide some insights into how search engineers view how search engines work. This one tells us that search ads located above search results are often referred to as “North” ads. Ads displayed below search results are often called “South” ads, and ads shown to the right of search results are “East” ads.
Getting back to the search category commercialization index, it isn’t uncommon that search queries are categorized into one or more categories. The author of the patent filing tells us that “this is especially common in search engines that double as Internet directories,” and that “the category or categories often reflect groupings of similar web destinations based on subject matter.”
Imagine if a search engine did provide some details on the effectiveness of categories of search terms used in advertising, to provide advertisers with an idea of current or potential commercialization of those categories of searches.
The Yahoo patent application introduces the idea of a search category commercialization index, which could be calculated by looking at two or more metrics from a group that would consist of:
- Number of searches within the category;
- Number of searches within the category that result in results pages having sponsored search results;
- Number of total advertisers in the category;
- Average number of sponsored search listings for results pages in the category
- Click-through rate for advertisements in the category;
- Share of sponsored search results that are clicked versus other items that are clicked on results pages in the category;
- Average price per click on a sponsored search result advertisement; and,
- Lifetime value of users who perform searches within the category.
The SCCI could be calculated based on two or more of those metrics and then may be compared to SCCIs for other categories in the search engine.
The patent application is:
Search Category Commercialization Index
Invented by Kavel Patel
Assigned to Yahoo
US Patent Application 20080133498
Published June 5, 2008
Filed December 5, 2006
Another set of insights from this patent filing, maybe more interesting than the use of compass points to describe the locations of ads, was this one:
During the inventive process surrounding the present invention, a significant amount of research was conducted to measure the usage patterns of users performing searchers, and specifically patterns involving their clicks on sponsored search advertisements. In this research, it was unexpectedly discovered that there are several metrics that may be monitored that aid in the determination of how well a category commercializes.
In this research, it was unexpectedly discovered that light users have the best likelihood of clicking on advertising, while searching for certain categories on the search results page. It was also unexpectedly discovered that a significant proportion of light user activity has no coverage.
In other words, it was unexpectedly discovered that a significantly large volume of search queries that resulted in pages with no coverage were potentially the most valuable type of search query. This realization resulted in approaching the problem from a completely new perspective–i.e., from the user perspective and how likely it is for a user’s behavior to turn to eventually reward the advertiser’s choice to advertise in the category by making some purchase.
Looking at Metrics that Could be Included in a Search Category Commercialization Index
Two or more of the following might be used. Of course, there may be others that just aren’t included in this patent filing.
Number of Searches within a Category — There may be a few different ways to measure this. One would involve simply looking at the total number of searches since the category was first created. Another would be to look at the total number of searches over a specific period of time, such as the past 6 months.
Limiting the time period may be helpful when comparing one category against another, since some categories may have been created long before or after others. And, since the tastes of users may change over time, for some categories, newer results may be more relevant. The author tells us:
For example, there may not have been a significant number of searches in the category of “MP3 Players” until iPods were released.
Coverage in a Category — Coverage refers to the number of searches that result in results pages having sponsored search results (i.e. ads based on the search terms). This could be the actual number of searches performed by all users, regardless of duplications. If thousands of searchers looked for “iPod,” which resulted in search pages having sponsored search results, then that would add thousands of positive results to the total number. This metric may also be expressed as a percentage of total searches in the category that resulted in results pages having sponsored search results.
The Amount of North Coverage — The compass points come in useful. These numbers could be included in the Coverage metric above, or may be viewed as a stand alone metric, looking at the number of searches in the category that result in results pages having sponsored search results at the top of the results pages.
Advertiser Participation in a Category — The number of total advertisers in the category. Of course, the more advertisers, the more commercial and competitive a category space is. While looking at these numbers can be helpful in deciding how commercial a category is, if there are a large number of advertisers, it is less likely that a searcher will purchase something from any one of them, because of the large amount of competition.
Average Number of Sponsored Search Listings within a Category — The number of ads typically showing up on a search page in a category.
Average Click-Through Rate for Advertisements in a Category. The click-through rate could be weighted based on the placement of the advertisements, since North advertisements are more likely to have higher click-though rates than South advertisements (interesting, but not surprising). So, this average click-through rate for a category may not be a true average but rather a weighted average with North advertisements having higher relevance.
Share of Sponsored Search Results that are Clicked On Versus Other Items Clicked On — Other items could be natural search results or links to related searches and other areas of the search engine.
Average Price Per Click on Sponsored Search Result Advertisements — We are told that this is a key metric for measuring true commercialization, because the average price per click can vary significantly between categories, such as Entertainment and Finance.
Lifetime value of the searcher — This value could be based on many different criteria. Is the user a long-time searcher, what is their frequency of searches, and do they follow seasonal patterns?
For example, a user who performs the majority of his searches at the end of the calendar year may very well be more valuable to advertisers than a user who spreads the searches out over the year, as it appears that the user may be more likely to be using searches to find items to purchase for Christmas.
Would providing advertisers with more information about how people search and interact with ads, and which categories of searches might provide better advertising opportunities inspire more advertising? It’s possible.
It’s difficult to tell how much information would actually be shared with advertisers, from the description of this Search Category Commercialization Index in the patent filing.